Parallel-process trajectories of alcohol use and bodily pain over 19 years: a Whitehall II study

Background

Preclinical evidence suggests a reciprocal relationship between alcohol use and chronic pain, but longitudinal co-development remains underexplored. Observational evidence among non-manual workers suggests a positive association, but most studies are cross-sectional. We identified co-developmental patterns of alcohol and pain among non-manual workers using joint growth mixture models (GMM) and parallel-process latent growth curve models (pp-LGM).

Methods

Data was analysed from five phases (5, 7, 9, 11, and 12; 1997-2016) of the Whitehall II cohort. Weekly alcohol consumption (1 unit=8g ethanol) was log2-transformed for normality. Pain was measured via the RAND-36 bodily pain score (MOS algorithm), inverted so higher values indicate greater severity. Joint GMM estimated trajectories with covariate adjustment; pp-LGM examined growth parameter correlations and cross-domain intercept-to-slope associations.

Findings

6,777 participants (mean age: 52.1 years; men: 67%) were included in primary analyses after removing deceased subjects. Four classes were identified. Classes 1-2 (29.7%) showed low (median<2 units), decreasing alcohol (-6.2% and -11.1%/phase) with either low-stable pain (intercept [i]: 19.6, slope [s]: 0.03 points/phase) or moderate-increasing pain (i: 49.8, s: 3.3). Class 3 showed a moderate (13.4 units), decreasing alcohol (-7.7%) with low-stable pain (i: 18.7, s: 0.4). Class 4 showed moderate (14.4 units), decreasing alcohol (-49.0%) with a low-increasing pain trajectory (i: 21.9, s: 1.3). The pp-LGM showed negative correlations between intercepts (r=-0.23, 95% CI: -0.27, -0.20) but not between slopes. Baseline alcohol predicted pain slope at borderline (β = 0.06, 95% CI: -0.01, 0.12).

Conclusions

Despite a declining trend in alcohol consumption and a rising trend in pain severity, the two trajectories showed no clear correlation. Alcohol may predict elevated pain severity, but this warrants validation through intervention-based approaches.

Key messages

• Alcohol use and pain were inversely correlated at baseline, but their rates of change were unrelated, so the opposing population trends do not reflect coupled trajectories.

• Baseline alcohol consumption might predict higher rate of change in pain severity, suggesting that alcohol reduction could warrant evaluation for pain prevention.